eHow
June 13, 2011
Research of human subjects or specimens involves observation and testing to describe a phenomenon. The things you wish to describe constitute the population. Because it is often unrealistic to test or observe large populations of people or specimens, researchers use sampling to make inferences about the population. A carefully selected sample represents the population in a study.
Sample vs. Population
If you wanted to conduct a study of how black teenage boys in Iowa handle stress, you might choose a sample of 100 black teenage boys at a Des Moines high school to administer stress tests and observe. It is unrealistic to test and observe every black teenage boy in the state. The sample, if carefully chosen, should represent the whole population of black teenage boys in Iowa. The sample's reactions to the stress test, in theory, would be the same as any group of black teenage boys in Iowa.
The population includes every single specimen that is like your sample. In the example, the population is every black teenage boy in Iowa. It would be impossible to test all members of the population. The value of sampling is that, if done correctly, it can tell a story about the population as a whole without the burden of observing a whole population. It is sometimes necessary to observe an entire population, such as when the population is too small to produce a statistically valid sample size.
Setting Parameters
Populations must be carefully described and samples carefully selected to ensure the reliability and validity of the study. Question rigorously the criteria used to select a population for sampling. How old are teenagers? How do you determine race? Is there something about the city or region being tested that makes it “typical”, or would a study across several regions give more representative results? Are there mitigating factors that could skew the results? Once a specific profile of the research population is described, sampling can begin.
Sampling Methods
Probability sampling is the best method to ensure reliable, valid results. Simple random samples (SRS) are most widely used because they yield representative, probabilistic results. In an SRS, every member of the population has an equal chance of selection for the sample. An example of SRS selection for a sample size of 100 would be to fill a box with all of the boys' names, shake it up and draw names randomly until 100 names have been drawn. Realistically, there are computer programs that will generate an SRS from the population. Other methods of probability sampling include stratified random sampling, systematic sampling and cluster sampling.
Bad Sampling Methods
Non-probability methods of sampling do not produce probabilistic samples and will not give generalizable results. One of these is availability or convenience sampling. For example: A researcher stands in a shopping mall and gives flyers to the people who walk by at noon on a Tuesday inviting them to fill out a survey online about stress. The problem with this method is that the sample may not be representative of the population in the region, or the population of shoppers who come to that mall. Shoppers in this example, who are not at work or school at noon on a weekday, may not reflect the population as a whole. They may have disproportionate disposable income, since they are at a mall. This method is easy, but can be damaging to a study. Other non-probability sampling methods include quota sampling, purposive sampling and snowball sampling.
References
College of the Sequoias: Population and Sample
Columbia Center for New Media Teaching and Learning: Types of Sampling
No comments:
Post a Comment